A1632
Title: Two-phase study design under differential unit costs: Insights from metaheuristic algorithms
Authors: Osvaldo Espin-Garcia - University of Western Ontario (Canada) [presenting]
Jiaqi Yuan - University of Western Ontario (Canada)
Abstract: Two-phase studies realize cost-efficiency by utilizing phase 1 data to select an informative subsample for expensive variable collection in phase 2. Previous work assumes equal cost of collecting expensive data across samples. This assumption is challenged and differential unit-cost two-phase designs are investigated. The proposed approach penalizes phase 2 subsamples based on budget constraint satisfaction. Three metaheuristic algorithms are compared under heterogeneous per-sample costs: Genetic Algorithm (GA), Simulated Annealing (SA), and Comprehensive Evolutionary Algorithm (CEA). The resulting designs aim to minimize the variance of the parameter of interest under the defined constraints. Simulations based on Poisson regression and inference via semiparametric maximum likelihood evaluate statistical performance across a range of design parameters. GA and CEA consistently achieve best performance while SA exhibits greater variability and occasional convergence failures.
